The present disclosure relates generally to data collection, and in particular to data collection based upon event detection for smart sensors on an arbitrating bus.
Condition monitoring systems are often used to monitor parameters in machinery. These systems include temperature sensors, pressure sensors, vibration sensors, and many other types of sensors, utilized to detect the development of faults. The data collected may be digitized, transformed and algorithmically processed to extract features of the monitored system. A given feature can be generated from the input of one sensor or could be dependent on the interaction of multiple sensors. Features extracted may be compared against expected values to assess the relative health of that feature. Sensor data may be monitored and evaluated on a scheduled basis and/or on an event basis. Event based monitoring is typically used to diagnose the circumstances leading to the event, identify the event, and diagnose the potential impacts of the event. Because of this, it is desirable to have data from all relevant sensors collected prior to, during, and after the event detection. This is typically accomplished by buffering data from sensors of interest, retaining the buffered data when an event is detected, and recording the additional post-event data.
Prior art condition monitoring systems included several analog sensors feeding into a central host for digitization and further processing. This implementation required individual cabling for each analog sensor which resulted in a heavy system that required a complicated installation. These issues can be mitigated by distributing digitization and processing of data out locally to the sensor. Local processing of data reduces the processing load on the host, allowing the host to perform other more complex functions, or decrease in size and complexity. Transmission of digital data allows multiple digital sensors to be connected on a single digital bus, reducing the wire weight associated with individual analog wire runs for each sensor.
In the prior art analog systems, the host was in control of all the data buffering as well as event detection, so data collection related to a detected event was relatively simple. With distributed processing, this data collection becomes more complex. It is desirable to develop methods of handling event detection and data collection for distributed processing sensor systems.